AI is already used in legal research and drafting, but no source we read expects it to replace lawyers. As of 24 October 2026, the US Bureau of Labor Statistics projects lawyer jobs to grow 5 percent from 2025 to 2035, while paralegal jobs are projected to stay flat. Bar guidance holds lawyers responsible for checking AI output, and courts have sanctioned lawyers who filed AI-invented cases. This page is general information, not legal advice.
01What the US job numbers say
The US Bureau of Labor Statistics (BLS) counted 863,700 lawyers in 2025 and projects growth of 5 percent to 2035. Its handbook says some routine legal work may be automated but that this efficiency is not expected to reduce overall demand for lawyers, because it frees time for client meetings and case strategy.
The outlook is weaker for support staff. BLS projects paralegals and legal assistants, 404,900 jobs in 2025, to see little or no change, and says technology including AI is expected to limit demand by making legal research and document preparation more efficient. In its 2025 case studies BLS named paralegals as the legal occupation likely to feel the strongest AI effect, and projected the legal services industry to grow 1.6 percent over 2023 to 2033 against 4.0 percent for the whole economy.
02What the American Bar Association says lawyers must do
In July 2024 the American Bar Association (ABA) issued Formal Opinion 512, its first ethics guidance on generative AI. In plain terms:
- Competence: understand what the specific tool can and cannot do. Becoming an AI expert is not required.
- Verification: do not rely on AI output without independent review. How much checking is enough depends on the tool and the task.
- Confidentiality: weigh the risk before entering client information, and in some cases get the client's informed consent, especially for tools that learn from inputs.
- Candor to the court: review citations and analysis before filing and fix fabricated cases or misstated law.
- Supervision and fees: firms need AI policies and training, and bills must reflect time actually spent.
The ABA expects this opinion to be followed by more specific guidance, from itself and from state and local bar committees, as the tools develop.
03When AI went wrong in court
These court orders are public. They concern lawyers' filings, not the tools' makers.
- Mata v. Avianca, Inc. (federal court, New York, 22 June 2023). Lawyers filed non-existent judicial opinions with fake quotes and citations produced by ChatGPT and kept defending them after the court raised doubts. The court said using a reliable AI tool is not improper in itself, but lawyers must still make sure filings are accurate. It found bad faith, imposed a $5,000 penalty and ordered letters to the client and to the judges falsely named.
- Johnson v. Dunn (federal court, Alabama, 23 July 2025). Two motions contained five problematic citations, which the record attributes to ChatGPT. The court found reckless conduct amounting to bad faith by the attorneys involved and ordered a public reprimand, removal from the rest of the case and referral to the state bar.
- Ayinde v Haringey and Al-Haroun v Qatar National Bank (High Court of England and Wales, 6 June 2025). The court said free generative AI tools cannot reliably do legal research and listed possible consequences, from wasted costs to regulator referral, contempt proceedings and, for deliberate falsehoods, police referral.
Researcher Damien Charlotin's public database listed 2,207 court decisions involving AI-made false content on 23 October 2026. It counts decisions that address the issue, not every fake citation.
04How reliable are legal AI tools?
A preregistered study published in the Journal of Empirical Legal Studies in 2025, and listed by Stanford Law School, tested research tools from LexisNexis and Thomson Reuters. It found that each hallucinated more than 17 percent of the time, though less often than the general chatbot GPT-4. Vendors update their tools, so the figures describe the versions tested.
If AI names a case or law, check it on an official court or legislation website before relying on it. This is general information, not legal advice. For your own situation, ask a licensed lawyer or your local legal aid or bar referral service.
?What is not known
- Courts and bar bodies have not settled one standard for when lawyers must tell clients or courts about AI use. The ABA says it depends on the circumstances.
- How often AI errors reach court filings is unknown. The Charlotin database counts only decisions that address the problem.
- How much AI will change hiring of junior lawyers and paralegals is not measured in the sources we found.
- Legal AI tools change quickly, so any test reflects only the versions that were tried.
QQuestions people ask
Can I use ChatGPT for legal research?
Courts have not banned AI use as such, but the English High Court said free tools cannot reliably do legal research. Anything an AI names must be checked against official sources, and this is general information, not legal advice.
Do lawyers have to tell clients they use AI?
ABA Opinion 512 says it depends on the circumstances. In some cases, particularly when client information goes into tools that learn from inputs, the lawyer needs the client's informed consent.
What happens to a lawyer who files AI-invented cases?
The orders above show penalties, public reprimand, removal from a case and referral to bar regulators. Outcomes depend on the court, the conduct and the country.
Last checked October 24, 2026. Rules and settings change, so open the source to see the current version.
- Lawyers, Occupational Outlook Handbook US Bureau of Labor Statistics Used for 2025 employment, projected growth to 2035 and the handbook's text on automation.
- Paralegals and Legal Assistants, Occupational Outlook Handbook US Bureau of Labor Statistics Used for 2025 employment, the flat projection and the text on AI limiting demand.
- Incorporating AI impacts in BLS employment projections: occupational case studies US Bureau of Labor Statistics (Monthly Labor Review) February 2025 article used for the legal occupation case study and industry growth figures.
- ABA issues first ethics guidance on a lawyer's use of AI tools American Bar Association News release on Formal Opinion 512, used for the duties of competence, confidentiality, candor, supervision and fees.
- Mata v. Avianca, Inc., No. 22-cv-1461, Opinion and Order on Sanctions (S.D.N.Y. June 22, 2023) US District Court, Southern District of New York The court's own order, used for the findings, the bad faith ruling and the sanctions.
- Johnson v. Dunn, 792 F. Supp. 3d 1241 (N.D. Ala. 2025) US District Court, Northern District of Alabama (copy hosted by Berkeley Law) Reported copy of the 23 July 2025 order, used for the facts and the sanctions imposed.
- Ayinde v London Borough of Haringey and Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin) High Court of England and Wales (Divisional Court), via the UK National Archives Judgment used for what the court said about AI research tools and the possible consequences.
- Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools Stanford Law School Publication page for the preregistered study used for the hallucination finding.
- AI Hallucination Cases Database Damien Charlotin (independent researcher) Public tracker used for the number of court decisions on AI-made false content and what it counts.
General information, not legal, medical, financial or investment advice. We are not affiliated with any company, agency or standards body named here. See our editorial policy and corrections, or tell us about a mistake at [email protected].